Ranking very low on your ROL

[Socrates25]

I know that theoretically the more places you rank, the better your chances at an interview.

But on closer examination, I'm not so sure if thats true for competitive specialties.

For example, lets say you rank 20 programs. For a very uncompetitive specialty, ranking #15-#20 probably increases your odds in a linear fashion of matching.

However, for a very competitive specialty, I dont think it works out like that. If you rank a program at #20, what are the odds that there are actually lesser matches out there that dont bump you out of a tentative match.

Here's a theoretical example (assuming a competitive subspecialty):

1. Program has 5 spots
2. Program ranks you #6
3. You rank program #20
4. Assume that you have failed to make a tentative match at a program higher on your ROL
5. The program ranks 50 people total

The only way you will match with a ROL of 20:1 is if the programs 49 other applicants have a greater match "agreement"; i.e. the distance between the applicant rank list and the program rank list is higher than 20:1.

I submit to you that the odds of distance > 20:1 in 49 other applicants with only 5 spots to fill a program would be EXTREMELY unlikely.

I guess my point is that in competitive specialties, ranking more than X number of programs wont do anything to increase your odds of matching. Now I'm sure X varies greatly between specialties. Am I thinking about this thing wrong?

Does anybody have stories about matching in a very competitive subspecialty with their #12 or #15 choice? I'd like to hear them.

---

Members don't see this ad.

 

[DrBowtie]

If you fall to spot 20 on your list, you get the spot since they ranked you #1.

I'm not sure about your question.

 

[Civil8]

If the program has ranked you #1, and you ranked them #20. If you haven't been able to match at your #1-19, then you will match at #20, because no one else can beat you out for that spot.

Have you read the NRMP's website on the match algorithm? It works exactly the same, no matter how competitive the specialty.

http://www.nrmp.org/res_match/about_res/algorithms.html

 

[Socrates25]

so a 20:1 match beats out a 2:2 match?

 

[gutonc]

I guess my point is that in competitive specialties, ranking more than X number of programs wont do anything to increase your odds of matching. Now I'm sure X varies greatly between specialties. Am I thinking about this thing wrong?

Yes you are.

The example you give (assuming I'm interpreting it correctly, which I may not because it makes almost no sense the way you explained it) is incorrect.

If you rank Program X #20 and they rank you #1, if you don't match to your #1-19, you WILL match at Program X, NO MATTER HOW MANY OTHER PEOPLE RANK THEM #1. In this scenario, by increasing the number of programs you rank from 10 to 20 you changed you chance of matching from 0 to 100%.

You've been through this once already, you should know how it works (and I think you actually do). But you're trying to make this infinitely more complicated than it is.

 

[aznsuperm3]

I dunno if the "match agreement" has anything to do with your odds of matching.

Correct me if I am wrong. My understanding of the algorithm is if applicant A rank a program #20 and applicant B ranks the same program #1, but the program ranks applicant A one rank higher than applicant B, applicant A will still get the spot over applicant B given that applicant A did not match at any other programs above #20. Hence the advice is always to rank as many programs as you are willing to attend.

 

[Socrates25]

OK then lets change the scenario to the program ranking you #6 instead of #1.

 

[gutonc]

so a 20:1 match beats out a 2:2 match?

Stop making up nonsense ways to describe a rank list. It makes it difficult to figure out what you're talking about.

 

[aznsuperm3]

OK then lets change the scenario to the program ranking you #6 instead of #1.

I think if they rank u #6 and rank the other applicant #7, you will get the spot over the other applicant regardless how the other applicant ranks the program given you did not match at the 19 other programs on your list.

 

[Socrates25]

Stop making up nonsense ways to describe a rank list. It makes it difficult to figure out what you're talking about.

I apologize sincerely for offending you in such a deep and personal manner.

 

[Socrates25]

I dunno if the "match agreement" has anything to do with your odds of matching.

Correct me if I am wrong. My understanding of the algorithm is if applicant A rank a program #20 and applicant B ranks the same program #1, but the program ranks applicant A one rank higher than applicant B, applicant A will still get the spot over applicant B given that applicant A did not match at any other programs above #20. Hence the advice is always to rank as many programs as you are willing to attend.

Yeah that makes sense, I'm sure that ranking 10 programs gives you a much better chance than ranking at 5 programs. I'm just not sure that ranking 30 programs gives you that much of a better chance than ranking 15. I'm sure there some very small increase in odds, but not so much that it warrants spending the extra money to go on all those interviews.

 

[PSurg530]

I don't understand why it wouldn't give you a better chance, especially depending on the quality/competitiveness of each program. Just like applying to more programs increases your chance of getting interviews. Yes, statistically the vast majority match in their top 3, but I would think especially for more competitive programs, you would have to go further down your list. If you look at the NRMP charting outcomes, they tell you the median number of contiguous ranks to match in different specialties, and it is longer for more competitive specialties.

I don't really know what you mean by 20:1 vs 2:2. You mean program is #20 on your list and you are #1 on theirs?

 

[2012mdc]

Yeah that makes sense, I'm sure that ranking 10 programs gives you a much better chance than ranking at 5 programs. I'm just not sure that ranking 30 programs gives you that much of a better chance than ranking 15. I'm sure there some very small increase in odds, but not so much that it warrants spending the extra money to go on all those interviews.

Just look at the charting outcomes for your specific specialty. There's a graph that shows where the plateau is in terms of more ranks not increasing your % chance of matching by much. Look at derm's chart on page 38. It plateaus around 18 ranks and there is a very minimal benefit after 15 ranks. If you look at plastics (pg 233) the graph is more linear and seems to keep rising past 20 ranks. So very specialty dependent even when comparing small super competitive specialties

www.nrmp.org/data/chartingoutcomes2011.pdf

 

[Civil8]

The match algorithm isn't nearly as complicated as you are attempting to make it, and it can in no way be expressed in terms of probability. I seriously suggest you read the NRMP's explanation it should address this completely. But here it goes anyway...

For matching in a competitive field, especially one with a small number of annual residents, extending your rank list will only increase your chances of matching. Statistically speaking, in most fields, the benefit will decrease at a certain point (see 2012mdc's post).

The most important thing to remember is that once you have been "beaten out" for a program by higher ranked applicants for your #1 rank, your #2 rank because your de-facto #1 rank. And so on, until you rank list ends, and you become unmatched.

 

[Dirt]

Just look at the charting outcomes for your specific specialty. There's a graph that shows where the plateau is in terms of more ranks not increasing your % chance of matching by much. Look at derm's chart on page 38. It plateaus around 18 ranks and there is a very minimal benefit after 15 ranks. If you look at plastics (pg 233) the graph is more linear and seems to keep rising past 20 ranks. So very specialty dependent even when comparing small super competitive specialties

www.nrmp.org/data/chartingoutcomes2011.pdf

People tend to misinterpret that data.

Just because last year, 80% of those who ranked X number of programs matched does not in any way mean that you have an 80% chance of matching if you rank that same number of programs. It does not answer the question of how many programs one needs to rank to match.

The point you people are trying to make is that adding more programs has diminishing returns but this is not really true. As Gutonc pointed out, if you rank 30 instead of 25 programs and you match at number 30 those extra 5 improved your chances from 0 to 100%.

 

[gutonc]

I apologize sincerely for offending you in such a deep and personal manner.

You didn't offend me but I can't help you if you're just making stuff up that makes no sense.

 

[gutonc]

OK then lets change the scenario to the program ranking you #6 instead of #1.

This obviously depends on what happens with the people they ranked 1-5 but if you move into "rank to match" territory by virtue of folks above you on a program's list matching elsewhere then we're back to the same situation as above.

Obviously there will be a point at which further interviews/ranks are unlikely to significantly improve your Match chances. The problem is that you don't (and can't), a priori know what that number is. It may be 5 or it may be 50. This is the point where "you pay your money and you take your chances" comes into play.

You're right...interviewing is expensive and even if you have unlimited financial resources, you're unlikely to have unlimited amounts of time to go on interviews so you have to make some decisions about how many and which interviews to go on. But if you're trying to create an equation that will help you figure out exactly how many programs you need to rank in order to guarantee a match, it's not going to happen, primarily because there are too many unknown variables (i.e. the rank lists of all the programs you interview at, all the rank lists of the other people who interview at the same programs you do, AND all the rank lists of the programs that interviewed those applicants, even if you didn't interview there).

Go on as many interviews as you can afford (time-wise and financially) and rank all the programs you could stand to be at. That's all you can really do here.

 

[mcl]

I think that Match is one of those phenomenon that doesn't benefit much from statistical analysis. Sure, match rates may not increase significantly (or even perceptively) when ranking between 15 and 20 programs--but if you are the person who matches at program #20 it's the difference between a secure spot and the Scramble. I'd hate for someone who interviewed at 20 programs to stop ranking them at 15 because they think it's pointless.

Rank all the programs you wouldn't mind matching to in the order in which you'd prefer to match and everything should be fine.

 

[PSurg530]

I think that Match is one of those phenomenon that doesn't benefit much from statistical analysis. Sure, match rates may not increase significantly (or even perceptively) when ranking between 15 and 20 programs--but if you are the person who matches at program #20 it's the difference between a secure spot and the Scramble. I'd hate for someone who interviewed at 20 programs to stop ranking them at 15 because they think it's pointless.

Rank all the programs you wouldn't mind matching to in the order in which you'd prefer to match and everything should be fine.

 

[2012mdc]

People tend to misinterpret that data.

Just because last year, 80% of those who ranked X number of programs matched does not in any way mean that you have an 80% chance of matching if you rank that same number of programs. It does not answer the question of how many programs one needs to rank to match.

The point you people are trying to make is that adding more programs has diminishing returns but this is not really true. As Gutonc pointed out, if you rank 30 instead of 25 programs and you match at number 30 those extra 5 improved your chances from 0 to 100%.

True. You can't exactly extrapolate the data to an individual basis.

But when looking at it as a whole, if 99.9% match with 25 ranks that means only 0.1% would have benefited by having more ranks. This means the case of someone benefiting from 30 ranks as opposed to 25 is very very rare.

No one can predict the future so past data is the best we can do. If 100% of people in the specialty match with 20 ranks it is probably unnecessary to rank 30. Of course you can't prove it but it's a reasonable conclusion based on the data we have.

 

[2012mdc]

I think that Match is one of those phenomenon that doesn't benefit much from statistical analysis. Sure, match rates may not increase significantly (or even perceptively) when ranking between 15 and 20 programs--but if you are the person who matches at program #20 it's the difference between a secure spot and the Scramble. I'd hate for someone who interviewed at 20 programs to stop ranking them at 15 because they think it's pointless.

Rank all the programs you wouldn't mind matching to in the order in which you'd prefer to match and everything should be fine.

I agree. You should rank every program you interview at unless you rather scramble than be at that program.

The data comes into play when deciding how many interviews one should go on. A line has to be drawn somewhere because of physical, financial, and time constraints and the data helps in figuring out where to draw the line.

 

[K31]

so a 20:1 match beats out a 2:2 match?

As I understand it, other applicants' rank lists have no impact on where you match. As long as you haven't matched at a place further up your rank list, and a program hasn't filled with applicants higher on their (the program's) list than you, you will match there.

 

[Law2Doc]

As I understand it, other applicants' rank lists have no impact on where you match. As long as you haven't matched at a place further up your rank list, and a program hasn't filled with applicants higher on their (the program's) list than you, you will match there.

Correct. You go down your own match lust until you find someplace that liked you more than the other people. That means if your number 20 liked you to match, and your numbers 1-19 didn't, you will end up with 20, even if other people put that one as their first choice. It's not a complicated system if you don't get dilly and start trying to "game" it.

Also I agree that statistics shouldn't really dictate your moves on the match. You can end up matching in your first choice, or fall all the way to your twentieth. There isn't a magic number where you can be confident to match. Sure most US seniors will get one of their top three, but that doesn't really mean you won't be scrambling if you list six. It's not a random distribution -- your credentials and interviewing skills, etc play a role.. The guy one program loves is probably the guy every program loves, and vice versa. So if you are the guy who I going to fall past 3 spots, you might be the guy who falls past 20 spots too. So it's a good idea to have interviewed broadly, and rank all the places you interviewed at that seem like better options than scrambling. Never say, well I statistically only need 6 to match so I will stop ranking after 7 (even though I interviewed at a dozen).

 

[HumbleMD]

Also I agree that statistics shouldn't really dictate your moves on the match.

In epi terms, the NRMP charting outcomes is problematic because it is cross-sectional data, which means one can't prove causation. Additionally, the data isn't granular enough to apply to single applicants. People ranking 15 continuous programs can look very differently. A top candidate could rank 15 and match at #1, or a horribly interviewing candidate could rank 15 and scramble.

I agree with the OPs point that there are likely diminishing returns. The stakes are too high to bank on "likely" though.

Rank in order of preference at every program you'd rather go to than scrambling. Wait until March. Easy enough.

 

[Buzz Me]

In epi terms, the NRMP charting outcomes is problematic because it is cross-sectional data, which means one can't prove causation. Additionally, the data isn't granular enough to apply to single applicants.

Exactly. Like whenever I try to explain to pre-meds that just because a given med school had x number of graduates match to y program, doesn't mean you have a good chance at going there as well.

 

[Law2Doc]

Exactly. Like whenever I try to explain to pre-meds that just because a given med school had x number of graduates match to y program, doesn't mean you have a good chance at going there as well.

Well, a lot of programs do base their opinion of a school on the residents they got from there in previous years. I've seen places where the PD was happy with residents from school X in the past couple of years and so yes he's absolutely going to go back to that well in subsequent years -- there is a reasonably good shot for students of school X at this program. I've seen the converse too, where a guy from school Y was a horrible resident, and that basically burned a bridge for school Y in subsequent years. None of this should impact ranking though.

 

[hot sauce]

I know that theoretically the more places you rank, the better your chances at an interview.

But on closer examination, I'm not so sure if thats true for competitive specialties.

For example, lets say you rank 20 programs. For a very uncompetitive specialty, ranking #15-#20 probably increases your odds in a linear fashion of matching.

However, for a very competitive specialty, I dont think it works out like that. If you rank a program at #20, what are the odds that there are actually lesser matches out there that dont bump you out of a tentative match.

Here's a theoretical example (assuming a competitive subspecialty):

1. Program has 5 spots
2. Program ranks you #6
3. You rank program #20
4. Assume that you have failed to make a tentative match at a program higher on your ROL
5. The program ranks 50 people total

The only way you will match with a ROL of 20:1 is if the programs 49 other applicants have a greater match "agreement"; i.e. the distance between the applicant rank list and the program rank list is higher than 20:1.

I submit to you that the odds of distance > 20:1 in 49 other applicants with only 5 spots to fill a program would be EXTREMELY unlikely.

I guess my point is that in competitive specialties, ranking more than X number of programs wont do anything to increase your odds of matching. Now I'm sure X varies greatly between specialties. Am I thinking about this thing wrong?

The way you are wording this is confusing to me so I may have misunderstood you. I think you are not understanding the match algorithm as there is no "distance" factor.

My understanding of your scenario is you ranked 20 programs but for whatever reason did not match at 1-19 (let's say they didn't rank you just to make it easy). Now you are down to #20 on your list which ranked you 6 out of the 50 people they interviewed and they have 5 spots.

You actually have a strong chance of matching here but it is not certain. If the program matched their top 5 applicants then you do not match there. Otherwise (if even one of those top 5 matched somewhere else) you will match there. It doesn't matter at all what the 44 applicants below on the program's rank list did they could have all ranked it #1. You are next on the list; they wanted you more than the other 44.

In general most programs do not match their top 5 applicants for 5 spots even in competitive fields. So in your scenario you will probably match to #20.

Now in general the probability of dropping to #20 on your rank list and being that highly ranked at the 20th program is fairly low. However, ranking more programs that you interviewed at will always increase your chances of matching (unless you have information that you are not supposed to have).

As your probability of matching approaches 1 ranking more programs will make a smaller difference. So, the number of programs that you need to rank depends on how competitive you are and how competitive the programs you are ranking are but there is no magic number.

 

[HumbleMD]

...So, the number of programs that you need to rank depends on how competitive you are and how competitive the programs you are ranking are but there is no magic number.

These are the 2 things that things that aren't shown in the reported "number of continuous ranked programs" statistic. A super top candidate in say, medicine, will likely rank 12 programs and match, but they will also be ranking top competitive programs. They could have gone to mid or lower tier programs and only rank 2 or 3 and match, but this doesn't often happen. A bottom-feeder candidate will also likely rank 12 programs and match, as long as there are some mid or lower tier programs. If they only ranked their reach schools, they could just as easily rank 12 and scramble.

Again, interpreting the NRMP statistics should be less "rank 12 programs to successfully match" but more "people who successfully match happen to rank 12 programs." The number I really wish they reported would be what # on their list people tend to match at by specialty. I was surprised to see that over 50% of all US seniors match at their #1 (there of course is self-selection and interview censoring).

 

[Law2Doc]

... I was surprised to see that over 50% of all US seniors match at their #1 (there of course is self-selection and interview censoring).

yeah, you can interpret this multiple ways, but it might just really be saying that some large chunk of the residencies out there are noncompetitive and go to the US folks who rank them highly.

 

[Buzz Me]

I was surprised to see that over 50% of all US seniors match at their #1 (there of course is self-selection and interview censoring).

I've always taken this to mean that applicants tend to apply to specialties and programs where they'll be competitive...e.g. the below-average applicant won't apply to Derm, nor will he/she apply to Medicine but rank super-competitive programs at the top of the list.